US20170323004A1

US20170323004A1 - Block classified term

Info

Publication number: US20170323004A1
Application number: US15/524,122
Authority: US
Inventors: Daniel Lau; Lewis Mackay; Daniel Timms
Original assignee: Longsand Ltd
Current assignee: Longsand Ltd
Priority date: 2014-11-27
Filing date: 2014-11-27
Publication date: 2017-11-09
Also published as: JP2017534128A; EP3224738A1; WO2016082877A1; US10902026B2; CN107077471A

Abstract

A class may be determined of a term from a database. The term may be blocked from being presented to a user, if the determined class does not include a permission for the user to view the term. The term may suggest a remainder of an incomplete query input by the user.

Description

BACKGROUND

Device or systems may provide a feature called autocomplete, or word completion. Autocomplete may involve the device or system predicting a word or phrase that the user wants to type in without the user actually typing it in completely. Manufacturers, vendors, and/or service providers are challenged to provide improved autocomplete technologies to better assist the user.

BRIEF DESCRIPTION OF THE DRAWINGS

The following detailed description references the drawings, wherein:

FIG. 1 is an example block diagram of a device to block a term from being presented to a user;

FIG. 2 is another example block diagram of a device to block a term from being presented to a user;

FIG. 3 is an example block diagram of a computing device including instructions for blocking a term based on a class of the term; and

FIG. 4 is an example flowchart of a method for blocking a term based on a class of the term.

DETAILED DESCRIPTION

Specific details are given in the following description to provide a thorough understanding of embodiments. However, it will be understood that embodiments may be practiced without these specific details. For example, systems may be shown in block diagrams in order not to obscure embodiments in unnecessary detail. In other instances, well-known processes, structures and techniques may be shown without unnecessary detail in order to avoid obscuring embodiments.
Auto-completion dialogues may provide a user with suggestions from fragments of input text. For example “capit” may be auto-completed to “capital” or “capitulate.” Auto-completion may be implemented through, for example, web browsers, e-mail programs, search engine interfaces, source code editors, database query tools, word processors, and command line interpreters.
Some implementations may use either a dictionary or search engine. The search engine may only provide suggestions that return relevant items indexed into the search engine, as opposed to a dictionary where some entries may not be present. However, in some scenarios the indexed data may include sensitive information. For example, a search index of medical records could contain patient names or their social security numbers. Auto-completing sensitive information may be undesirable whilst completing non-sensitive information is beneficial to the search operator.
Filtering data using only weighting or some popularity/threshold parameter (number of documents containing terms), may not provide fine enough control to prevent leaking of sensitive information. Further, providing explicit blacklists for suggestions may filter out exact term matches. However, manually providing and/or updating such a level of fine control may be cost-prohibitive, to the point where it is unlikely to be usefully applied.
Examples may use classification technology to filter auto-complete suggestions so that users are presented only with information they are permitted to see. An example device may determine a class a term from a database. The device may block the term from being presented to a user, if the determined class does not include a permission for the user to view the term. The term may suggest a remainder of an incomplete query input by the user.
Thus, examples may allow for finer control over what elements are filtered compared to simple weight/threshold parameters. Further, examples may allow for faster deployment and less maintenance compared to a manually maintained blacklist or whitelist of exact terms/phrases/entries.
Referring now to the drawings, FIG. 1 is an example block diagram of a device 100 to block a term from being presented to a user. The device 100 may be a microprocessor, a controller, a memory module or device, a notebook computer, a desktop computer, an all-in-one system, a server, a network device, a wireless device, or any other type of device capable of interacting with a database and/or intercepting a message along a network.
The device 100 is shown to include a classification unit 110 and a filter unit 120. The classification and filter units 110 and 120 may include, for example, a hardware device including electronic circuitry for implementing the functionality described below, such as control logic and/or memory. In addition or as an alternative, the classification and filter units 110 and 120 may be implemented as a series of instructions encoded on a machine-readable storage medium and executable by a processor.
The classification unit 110 may determine a class 112 of a term from a database. The term may be a word or phrase used to describe a thing or to express a concept, such as a name, an address, and a social security number, and the like. The term may suggest a remainder of an incomplete query input by the user. The class 112 may relate to a system for identifying various types of terms, such as confidential and non-confidential terms.
The filter unit 120 may block a term from being presented to a user, if the determined class 112 does not include a permission 122 for the user to view the term. For instance, the determined class 112 may indicate at least one of sensitive and personally identifiable information, if the determined class 112 does not include permission 122 for the user to view the term. The filter unit 120 may allow the term to be presented to the user, if the determined class 112 includes the permission 122 for the user to view the term.
The user may be any person who is entering a query, such as by using a computer or network service, for which the database may autocomplete with the term. The user may have a user account and/or be identified by a user name and/or password. The permission 122 may relate to the whether the user has a right to view, access or modify the term. The permission 122 here may relate to whether the user may view the term triggered by the database in response to the user's query.
For instance, if the user does not have permission to view the term based on the class 112 of the term, the filter unit 120 may block the term by preventing the term from being sent to the user and/or denying access to the term. The determined class 112 may be stored and/or associated with the term at the database, the classification unit 110 and/or the filter unit 120, such as via metadata. The classification and filter units 110 and 120 are explained in greater detail below with respect to FIG. 2.
FIG. 2 is another example block diagram of a device 200 to block a term from being presented to a user. The device 200 may be a microprocessor, a controller, a memory module or device, a notebook computer, a desktop computer, an all-in-one system, a server, a network device, a wireless device, or any other type of device capable of interacting with a database and/or intercepting a message along a network.
The device 200 is shown to interface with a database 230. The database 230 may be any electronic, magnetic, optical, or other physical storage device that contains or stores information, such as Random Access Memory (RAM), an Electrically Erasable Programmable Read-Only Memory (EEPROM), a storage drive, a Compact Disc Read Only Memory (CD-ROM), and the like. For instance, the database 230 may include the most popular search terms 232-1 to 232-n, where n is a natural number, indexed from a search engine. Further, at least some of the search terms 232-1 to 232-n, may include personally identifiable information (PII), such as medical records, names, social security numbers and the like.
The device 200 of FIG. 2 may include at least the functionality and/or hardware of the device 100 of FIG. 1. For example, a classification unit 210 of the device 200 of FIG. 2 may include at least the functionality and/or hardware of the classification unit 110 of the device 100 of FIG. 1 and a filter unit 220 of the device 200 of FIG. 2 may include at least the functionality and/or hardware of the classification unit 120 of the device 100 of FIG. 1.
As noted above, the classification unit 210 may determine a class 212 of a term 232 from the database 230. The class 212 of the term 232 may vary with respect to the user 250. For example, the term 232 may be classified as confidential with respect to a first user but classified as non-confidential with respect to a second user. Thus, the classification unit 210 may take into account a type or identify of the user 250 when determining the class 212 of the term 232. Different types of the users 250 may correspond to different types of classes 212. For instance, the user's 250 account may be used to identify the type of user, such as when the user 250 logs into a system.
As also noted above, the filter unit 220 may block a term from being presented to a user 250, if the determined class 212 does not include a permission 222 for the user 250 to view the term 232. The filter unit 220 may allow the term 232 to be presented to the user 250, if the determined class 212 includes the permission 222 for the user 250 to view the term 232.
The classification unit 210 may classify the term 232 based on at least one of a rule 214 and machine learning 216. While one rule 214 is shown, examples may include a plurality of rules. The rule 214 may indicate an operation to be performed on a number, letter, grammar, punctuation and/or syntax of the term 232. The classification unit 210 may use the rule 214 to match the term 232 to at least one of a template and a pattern. For example, the classification unit 210 may use a rule to classify a term 232 as a social security number, if the term 232 matches a particular pattern for a social security number, as indicated by the rule 214. The filter unit 220 may block the term 232 from being presented to the user 250, if the term 232 is classified as a social security number.
In another example, the classification unit 210 may perform an arithmetic operation on the term 232. In turn, the filter unit 220 may allow the term to be presented to the user 250, if a result of the arithmetic operation satisfies the rule 214. For instance, the classification unit 210 may classify the term 232 as a credit card number upon a result of a checksum or multiplication of the digits of the credit card or instead classify the term 232 as a date upon comparing a range and/or syntax of the term 232 to a template. Here, the filter unit 220 may block the term 232 from being presented to the user 250, if the term 232 is classified as a credit card number or a date that falls on prohibited day.
Machine learning 216 may relate to a construction and study of algorithms that can learn from data. Such algorithms may operate by building a model based on inputs and using that to make predictions or decisions, rather than following only explicitly programmed instructions. Machine learning 216 techniques may include, for example, grammar induction and/or a probabilistic classifier. For instance, the probabilistic classifier may be a Bayesian classifier. Grammar induction may include, for example, inference by trial-and-error, a genetic algorithm, a greedy algorithm, a distributional learning algorithm and a pattern learning algorithm. The classification unit 210 may use machine learning to classify types of terms 232 that may not be easily identifiable via a rule 214, such as addresses or spam.
As noted above, the classification unit 210 may determine a plurality of the different types of classes 212, based on the plurality of terms 232-1 to 232-n included in the database 230. The types of classes 212 may relate to different security clearances. Further, at least one of the classes 212 may be a subset of another of the classes 212. Thus, the filter unit 220 may compare to an identify of the user 250 to class 212 of the term 232 determine, if the user's security clearance only allows them to see a subset of the terms 232. If the user 240 does have not security clearance, the filer unit 220 may not provide the term 232 to the user 250, which was suggested by the database in response to the user's 250 query.
The classification unit 210 may determine a plurality of the classes 212 of the terms 232 simultaneously. Similarly, the filter unit 220 may block and/or allow a plurality of the terms 232 simultaneously. Thus, examples may remove or prevent terms 232 from being suggested to the user 250 that are classified as not to be presented to the user 250. Further, P11 is just one example of a type classification that could be filtered upon by the filter unit 220. Examples may determine a class 212 of a term 232, based on any type of criteria deemed appropriate for denying to the term 232.
FIG. 3 is an example block diagram of a computing device 300 including instructions for blocking a term based on a class of the term. In the embodiment of FIG. 3, the computing device 300 includes a processor 310 and a machine-readable storage medium 320. The machine-readable storage medium 320 further includes instructions 322 and 324 for blocking the term based on the class of the term.
The computing device 300 may be included in or part of, for example, a microprocessor, a controller, a memory module or device, a notebook computer, a desktop computer, an all-in-one system, a server, a network device, a wireless device, or any other type of device capable of executing the instructions 322 and 324. In certain examples, the computing device 300 may include or be connected to additional components such as memories, controllers, etc.
The processor 310 may be, at least one central processing unit (CPU), at least one semiconductor-based microprocessor, at least one graphics processing unit (GPU), a microcontroller, special purpose logic hardware controlled by microcode or other hardware devices suitable for retrieval and execution of instructions stored in the machine-readable storage medium 320, or combinations thereof. The processor 310 may fetch, decode, and execute instructions 321, 323, 325, 327 and 329 to implement blocking the term based on the class of the term. As an alternative or in addition to retrieving and executing instructions, the processor 310 may include at least one integrated circuit (IC), other control logic, other electronic circuits, or combinations thereof that include a number of electronic components for performing the functionality of instructions 322 and 324.
The machine-readable storage medium 320 may be any electronic, magnetic, optical, or other physical storage device that contains or stores executable instructions. Thus, the machine-readable storage medium 320 may be, for example, Random Access Memory (RAM), an Electrically Erasable Programmable Read-Only Memory (EEPROM), a storage drive, a Compact Disc Read Only Memory (CD-ROM), and the like. As such, the machine-readable storage medium 320 can be non-transitory. As described in detail below, machine-readable storage medium 320 may be encoded with a series of executable instructions for blocking the term based on the class of the term.
Moreover, the instructions 322 and 324, when executed by a processor (e.g., via one processing element or multiple processing elements of the processor) can cause the processor to perform processes, such as, the process of FIG. 4. For example, the analyze instructions 322 may be executed by the processor 310 to analyze a term from a database (not shown) to determine a class, the term is to relate to part of a query and to suggest a remainder of the query. The determine instructions 324 may be executed by the processor 310 to determine if the term is to be blocked in response to the query, based on the class of the analyzed term. The class may be determined based on at least one of a rule and machine learning. For example, the term may be blocked from being presented, if a user does not have permission to the analyzed class. The term may be allowed to be presented, if the user, if the user has permission to the analyzed class.
FIG. 4 is an example flowchart 400 of a method for blocking a term based on a class of the term. Although execution of the method 400 is described below with reference to the device 200, other suitable components for execution of the method 400 can be utilized, such as the device 100. Additionally, the components for executing the method 400 may be spread among multiple devices (e.g., a processing device in communication with input and output devices). In certain scenarios, multiple devices acting in coordination can be considered a single device to perform the method 400. The method 400 may be implemented in the form of executable instructions stored on a machine-readable storage medium, such as storage medium 320, and/or in the form of electronic circuitry.
At block 410, the device 200 receives a term 232 from a database 230 related to part of a query of a user 250. The term 232 may suggest a remainder of the query. At block 420, the device 200 may classify the term based on at least one of a rule 214 and machine learning 216. The machine learning 216 may include at least one of grammar induction and a probabilistic classifier to classify the term 232. The rule 214 may match the term to at least one of a template and a pattern to classify the term 232.
At block 430, the device 200 blocks the term 232 from being suggested, if the class 212 of the term 232 does not provide permission 222 to a user 250 to view the term 232. At block 440, the device 200 allows the term to be suggested, if the class 212 of the term 232 does provide permission 222 to the user 250 to view the term 232.

Claims

We claim:

1. A device, comprising:

a classification unit to determine a class of a term from a database; and

a filter unit to block the term from being presented to a user, if the determined class does not include a permission for the user to view the term, wherein

the term is to suggest a remainder of an incomplete query input by the user.

2. The device of claim 1, wherein the classification unit is to classify the term based on at least one of a rule and machine learning.

3. The device of claim 2, wherein,

the classification unit is classify the term based on machine learning, and

the machine learning includes at least one of grammar induction and a probabilistic classifier.

4. The device of claim 3, wherein,

the probabilistic classifier includes a Bayesian classifier; and

the grammar induction includes at least one of inference by trial-and-error, a genetic algorithm, a greedy algorithm, a distributional learning algorithm and a pattern learning algorithm.

5. The device of claim 2, wherein

the classification unit is to classify the term based on the rule, and

the rule indicates an operation to be performed on at least one of a number, letters and syntax of the term, and

the classification unit is to use the rule to match the term to at least one of a template and a pattern.

6. The device of claim 5, wherein,

the classification unit is to perform an arithmetic operation on the term, and

the filter unit is to allow the term to be presented to the user, if a result of the arithmetic operation satisfies the rule.

7. The device of claim 1, wherein,

the filter unit is to allow the term to be presented to the user, if the determined class includes the permission for the user to view the term, and

different types of the users correspond to different types of classes.

8. The device of claim 7, wherein,

the classification unit is determine a plurality of the different types of classes, based on the plurality of terms included in the database, and

the terms of the database are mined from data indexed into a search engine.

9. The device of claim 8, wherein,

the types of classes relates to different security clearances, and

at least one of the classes is a subset of another of the classes.

10. The device of claim 1, wherein,

the term includes at least one of a name, an address, and a social security number, and

the determined class indicates at least one of sensitive and personally identifiable information, if the determined class does not include permission for the user to view the term.

11. The device of claim 1, wherein

the classification unit is to determine a plurality of the classes of the terms simultaneously, and

the filter unit is to at least one block and allow a plurality of the terms simultaneously.

12. A method, comprising:

receiving a term from a database related to part of a query of a user, the term to suggest a remainder of the query;

classifying the term based on at least one of a rule and machine learning;

blocking the term from being suggested, if the class of the term does not provide permission to a user to view the term; and

allowing the term to be suggested, if the class of the term does provide permission to the user to view the term.

13. The method of claim 12, wherein,

the machine learning includes at least one of grammar induction and a probabilistic classifier to classify the term, and

the rule is to match the term to at least one of a template and a pattern to classify the term.

14. A non-transitory computer-readable storage medium storing instructions that, if executed by a processor of a device, cause the processor to:

analyze a term from a database to determine a class, the term is to relate to part of a query and to suggest a remainder of the query; and

determine if the term is to be blocked in response to the query, based on the class of the analyzed term, wherein,

the class is determined based on at least one of a rule and machine learning.

15. The non-transitory computer-readable storage medium of claim 14, wherein,

the term is blocked from being presented, if a user does not have permission to the analyzed class, and

the term is allowed to be presented, if the user has permission to the analyzed class.